Research And Control Of Key Quality Characteristics For Complex Electromechanical Products

With the development of production,complex mechanical and electrical products appear in various production and manufacturing processes,so the research on complex mechanical and electrical products is more and more important.The complex electromechanical products are characterized by complex structure,large quantity of quality characteristics and so on.The difficulty coefficient of production control and detection is large.Extracting key quality characteristics in production management and monitoring them can effectively guarantee product quality.In order to improve the accuracy extraction of critical-to-quality characteristics(CTQs)of complex electromechanical products and study the influence of the quantity of CTQs on quality control,the method of BP-MTS for CTQs extraction is constructed which combines the Martin system with the BP neural network.Average running length(ARL)is used as the evaluation index for the control performance of Hotelling T2 control chart.The effect of the quantity of CTQs on Hotelling T2 control graph is analyzed,and quality loss under different quantity of CTQs is calculated.Taking the cylinder of the engine as an example,a comparative analysis is made with the traditional Mazda system.The results show that:(l)the corresponding quantity of CTQs can be given according to different production needs by BP-MTS method;(2)if the mean shift is small,fewer CTQs should be selected for better production monitoring and less quality loss,otherwise,more CTQs are in favor of effective monitoring.According to the daily production control of complex electromechanical products,the corresponding control chart is analyzed.The Q statistics were combined with the EWMA control chart,and the RSS-QEWMA control chart was improved and constructed by using the ordered Set Sampling(RSS)to improve the monitoring performance of the EWMA control chart under different quantity and quality characteristics of complex electromechanical products.The Average running chain length(ARL)and Average Time to Signal(ATS)were used to analyze the performance of control charts,and the monitoring performance was compared with that of EWMA,CUSUM and Hotelling T2 control charts.The RSS-QEWMA has better monitoring performance than the EWMA and Hotelling T2 control charts when the quality characteristic quantity(p=1)and the process mean offset(δ≤0.7);when it is p=1 δ>0.7,the monitoring effect is better than the CUSUM control chart;at that time,the RSS-QEWMA control chart can maintain good monitoring performance when p>1.In order to realize the economic statistics of MEWMA(Multivariate Exponential Weighted Moving average)control graph under specific production conditions,an rss-mewma control graph based on the rank Set Sampling(RSS)was constructed.The corresponding cost model is established and the secondary loss function of Taguchi is used to correct it.The statistical performance of the control graph was analyzed,and the calculation method of Average Time to Signal(ATS)and Average Number of Observations to Signal(ANOS)based on markov chain was constructed.ATS and ANOS were taken as evaluation indicators when the process was out of control and compared with the economic and statistical characteristics of the traditional MEWMA control graph.It is proved that rss-mewma control graph based on economic statistical analysis can not only identify the deviation in the production process for alarm,but also reduce the corresponding control Cost.